12 projects
cendor-core
Wrap your LLM client once and capture exact token counts and cost on every call — the shared foundation the other Cendor tools build on.
cendor-tokenguard
Spending limits for LLM calls: stop a call before it runs if it would blow the budget, and see cost broken down per feature.
cendor-squeeze
Shrink long prompts to hit a token budget and expand them back exactly — the same input always produces the same output.
cendor-sdk
Build an LLM agent with spending limits, a tamper-evident audit trail, PII redaction, and record/replay testing built in from the start — a governed agent in about 10 lines.
cendor-contextkit
Fit a prompt into the model's token budget — add your messages and context, pin what must stay, and the least-important parts are dropped to make it fit.
cendor-mcp
Read-only MCP server that hands your AI coding assistant Cendor's live docs and correct call-shapes. Run offline with `uvx cendor-mcp`.
cendor-guardrails
A local-first gate for LLM apps: define a check — keyword, regex, URL, length, JSON-schema — attach it to a stage (input, tool call, tool output, output). Deterministic, microsecond, $0, and every decision is audit-grade evidence.
cendor-acttrace
An automatic, tamper-evident log of your AI's decisions — every model and tool call is recorded and can be verified offline; change any past entry and the check fails.
cendor-init
Offline CLI to make your project Cendor-ready and Cendor-fluent for your AI assistant (init) and catch wiring mistakes (doctor). Run with `uvx cendor-init`.
cendor-cassette
Record real LLM calls once, then replay them in your tests forever — offline, no API key, no cost, and the same result every run.
cendor-libs
All seven Cendor libraries in one install — plain building blocks for LLM apps: context, cost, compression, a deterministic guardrails gate, testing, and a tamper-evident audit trail.
cendor
Brand alias for the Cendor stack — installs cendor-libs (the full suite). Prefer `pip install cendor-libs`.
